主题：葫芦僧乱判葫芦案 -- 煮酒正熟

To be able to conclude whether a difference is statistically significant, you need a defined alpha in the first place. Normally alpha is defined at a level no greater than 5%, or 0.05. Alpha is also referred to as "confidence level". In essence, alpha is the level of risk that you erroneously reject the null hypothesis when it actually holds true. In other words, back to your point, the difference that you find is actually attributed to noise, but you conclude that the difference is due to a factor's effect, by mistake.

So as long as your P-value is below your pre-defined alpha, you can reject the null hypothesis and conclude that there IS a statistically significant difference, at x% confidence level. Note that by such a conclusion, you are not saying that you'll be absolutely correct in drawing that conclusion. Rather, you make a statement that you are correct under certain circumstances --- at a confidence level of a certain percentage.

Since our P-value is at the sixth or seventh digit left to the decimal position, way lower than 5% or 1% or even 0.1%, the difference's got to be the true difference, not the noise.

我不太明白的是，为什么您的实验结果是以差距的百分比为判断依据，而不是P-value作为判断依据？